Published in the May 2004 issue of Today’s Hospitalist
While electronic prescribing is generally regarded as a medical informatics success story, less than 20 percent of U.S. physicians appear to be actually using those tools.
Companies like ePocrates allow physicians to download e-prescribing tools free of charge from the Internet, leading many analysts to believe that e-prescribing had been widely embraced by physicians. But a report released last month by the Washington-based eHealth Initiative has reached a different conclusion.
The report says that current surveys show that between 5 percent and 18 percent of physicians are thought to use e-prescribing tools. While this number represents an increase over the past three to five years, the report says, it is far from qualifying e-prescribing as standard practice.
The eHealth Initiative report identifies barriers to the universal adoption of e-prescribing tools. Challenges include startup costs and concerns that the tools will reduce physician efficiency, at least initially.
The report concludes that greater physician acceptance of e-prescribing tools will happen only when systems become easier to install, learn and use. It also says that vendors need to agree on standards so that different systems can work together.
More than 3 billion prescriptions are written each year. Studies have suggested that the widespread adoption of e-prescribing systems could save as much as $27 billion.
Last year, the eHealth Initiative launched an effort to encourage physicians to adopt e-prescribing tools. The report is online.
Can computerized order entry detect dangerous drug errors in the hospital?
While computerized order entry systems do a good job of catching drug errors, recent research says that many of the most serious errors may still slip through.
Researchers from Northwestern Memorial Hospital scrutinized the drug orders that came through a 700-bed Chicago teaching hospital. The orders were all placed in a one-week period in early 2002. The study appeared in the April 12, 2004, issue of the Archives of Internal Medicine.
Pharmacists sorted through more than 17,000 inpatient and emergency department drug orders to look for errors and determine which would have been caught by a computerized order entry system.
Researchers found that slightly more than 6.2 percent of the drug orders contained a confirmed prescribing error. Almost two-thirds of those errors occurred on the day of admission.
Most of the errors “69.2 percent “were categorized as “unlikely to have caused harm.” Nearly one-fifth “19.3 percent “were classified as “likely to have required monitoring,” and the remaining 11.5 percent were labeled “likely to have produced patient harm.”
Of the errors that would have caused some patient harm, dosing errors were the most common. They accounted for more than one-third of clinically significant errors.
In terms of drug types, anti-infective agents, cardiovascular agents and opioid analgesics accounted for 57 percent of significant errors.
Researchers determined that nearly two-thirds of all errors could likely have been prevented with a computerized order entry system. They found that about 13 percent would have likely slipped through an order entry system, and that just over 22 percent could possibly have been prevented.
When the errors were divided into clinically significant and insignificant categories, however, the picture changed. While researchers found that nearly 80 percent of insignificant errors could have been detected by an order entry system, they said that only about half of the more serious errors would have likely been caught.
The study notes that basic order entry systems without decision-support features are likely to catch even fewer serious errors. A system that simply warns a physician about a wrong drug order, the researchers conclude, will not prevent as many errors as a system that suggests an alternative or gives some parameters for ordering another drug.
The study also notes that while previous research has shown that implementing a homegrown order entry system can reduce serious drug errors by up to 80 percent, implementing a much more basic system has been shown to reduce prescribing errors by as little as 19 percent. The study is online.
Advisory report raises questions about diagnosis codes
Proposed recommendations from a presidential advisory committee are raising questions about what coding system the government will recommend for health information systems.
In April, the President’s Information Technology Advisory Committee released a report urging the government to do more to encourage U.S. health care organizations to adopt technology. The report calls on the federal government to support health care information technology through expanding demonstration projects and increasing reimbursement for providers who use technology.
Some of its more specific recommendations, however, raise questions about coding sets. One of its main recommendations called for a cost-benefit study of coding systems. The committee specifically said the government should explore using Snomed Clinical Terms as an alternative to the ICD-10-CM code set.
Analysts say that because the Snomed codes contain more detail, they offer a more complete picture of the services provided to patients. They say that when using the current ICD coding system, for example, it can be difficult to estimate duration of illness strictly by looking at diagnosis codes.
Most analysts also agree, however, that the Snomed codes could not completely replace ICD-9 codes for reimbursement purposes. They say that the Snomed coding system offers so many details that adjudicating a claim using only its codes would be difficult.
The issue is important to physicians and health care organizations that are considering purchasing medical records systems. These systems will need to use the code sets that the government recommends.
Currently, no one knows exactly which coding systems the government will settle on, leaving vendors and potential purchasers of medical records systems in limbo.
The committee’s draft recommendations are online. Its final report is expected in June.